The Great Lakes Entomologist

Volume 32 Numbers 1 & 2 - Spring/Summer 1999 Numbers Article 9 1 & 2 - Spring/Summer 1999

April 1999

Road Mortality of Dragonflies (Odonata) in a Great Lakes Coastal Wetland

Samuel Keith Riffell Michigan State University

Follow this and additional works at: https://scholar.valpo.edu/tgle

Part of the Entomology Commons

Recommended Citation Riffell, Samuel Keith 1999. "Road Mortality of Dragonflies (Odonata) in a Great Lakes Coastal Wetland," The Great Lakes Entomologist, vol 32 (1) Available at: https://scholar.valpo.edu/tgle/vol32/iss1/9

This Peer-Review Article is brought to you for free and open access by the Department of Biology at ValpoScholar. It has been accepted for inclusion in The Great Lakes Entomologist by an authorized administrator of ValpoScholar. For more information, please contact a ValpoScholar staff member at [email protected]. Riffell: Road Mortality of Dragonflies (Odonata) in a Great Lakes Coastal

1999 THE GREAT LAKES ENTOMOLOGIST 63

ROAD MORTALITY OF DRAGONFLIES (ODONATAJ IN A GREAT LAKES COASTAL WETLAND

Samuel Keith Riffell 1

ABSTRACT Although road mortality of vertebrates has been well studied, road mortal­ ity of invertebrates has rarely been studied or considered in management scenarios. Mackinac Bay is an extensive coastal wetland in northern Michi­ gan. It is bordered by a two-lane paved highway that separates the marsh, where dragonflies defend territories and breed, from the adjacent forest where dragonflies forage and rest. During mid-summer of 1997, daily collec­ tions of dragonfly corpses from the road and road edge were used to estimate daily mortality rates and sex ratios among casualties. Daily mortality was highly variable, ranging from 10 to 256 casualties per kilometer. Sex ratios among casualties were generally male-skewed (60% or higher). Life-history differences between the sexes present a parsimonious explanation for male­ specific mortality. Mortality was even or female-skewed for some species, and impacts of road mortality may be more severe in populations where mortality is female-skewed. More research about the effects of roads on dragonflies is warranted because dragonfly populations are small relative to many inverte­ brates and are restricted to wetland habitats which are being degraded or de­ stroyed in many regions.

Wildlife mortality from collisions with motor vehicles along roads is a com­ mon phenomenon and a serious management concern. In some regions, road­ kills are the primary source of mortality for large mammals (Lalo 1987, Har­ ris and Gallagher 1989), and road-kills can be a substantial source of mortality for birds (Forman 1995, Ashley and Robinson 1996), amphibians (e.g. van Gelder 1973, Fahrig et al. 1995, Ashley and Robinson 1996), and reptiles (Rosen and Lowe 1994). Over time, road mortality can create geneti­ cally disjunct subpopulations (Merriam et al. 1989), depress population sizes (Rosen and Lowe 1994, Fahrig et al. 1995), and alter the size- and age-struc­ ture of populations (Rosen and Lowe 1994). Consequently, road mortality data have been incorporated into predictive models (e.g. grizzly bear; Doak 1995), and management schemes (e.g. road underpasses for Florida pan­ thers; Harris and Gallagher 1989) for many vertebrate species. However, even the most basic data about road mortality is exceedingly scant for aquatic invertebrates as well as terrestrial invertebrates (see Table 1 for a summary) despite anecdotal evidence that extremely higb numbers of inver­ tebrates can be killed on roads. The importance of road mortality in aquatic invertebrates has most likely been ignored because of their life-history traits. Most groups of aquatic in­

lDepartment of Zoology, Michigan State University, East Lansing, MI 48824.

Published by ValpoScholar, 1999 1 The Great Lakes Entomologist, Vol. 32, No. 1 [1999], Art. 9

0­ ~

Table 1. Road mortality rates for invertebrates from other studies with selected rates from this study for comparison. Invertebrate Taxon Study Duration Events # Road Length Mortality Rate PUBLISHED MORTALITY RATES Beckemeyer ....., I Odonata: A. junius 1 day 1 9 ~ 0.5 km 45.00 f km (day" m G) ;;0 Seibert and Conover (1991) m All Insecta 14 months 50 1069 1.6km 13.36(kmf ~ All Odonata 14 months 50 22 1.6km 0.28/km ( s;: ?\ m Munguira and Thomas (1992) (J> All Bu Lterflies 44 days 4 144 1.2km 30.00 (km (day" m Z DRAGONFLY MORTALITY RATES AT MACKINAC BAY b s: All Odonata 42 days 26 1140 0.5 km 87.691 km 1day o L. quadrimaculata 42 days 26 312 0.5 km 24.00 Ikm I day 5 E..~pingera 42 days 26 199 0.5 km 15.311 km (day G) .....,Vi aConverted from an estimated 72 per mile of 4 lane highway based on sampling - 0.5 km of a single lane (see Beckemeyer 1996). bCalculated assuming each of the 50 sampling events represents 24 hours ofmortality. cCak'Ulated assuming each of the 4 sampling events represents 24 hours of mortality. However, Munguira and Thomas (1992) demon­ strated that corpses did not disappear after 2 weeks time, and assumed the 144 carcasses represented the entire 44 days of mortality. ~ Using this assumption, the mortality rate would be much lower (2.72 km I day). W I'V z ::>

9<> I'V https://scholar.valpo.edu/tgle/vol32/iss1/9 2 ==-___= __iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii ______,.-_

Riffell: Road Mortality of Dragonflies (Odonata) in a Great Lakes Coastal

1999 THE GREAT LAKES ENTOMOLOGIST 65

sects with winged adult stages are highly fecund with naturally high mortal­ ity (Wallace and Anderson 1996), and have a synchronized emergence of short-lived adults which fly primarily at night (Trichoptera caddisflies; Wig­ gins 1996) or breed in mass swarms (e.g. Ephemeroptera mayflies; Edmunds and Waltz 1996). A common assumption is that road mortality of adults has little potential to impact population dynamics of a species with this combina­ tion of life-history traits. This assumption, however, has not been tested, and not all groups of aquatic insects (i.e. Odonata) share these traits. Dragonflies differ from many groups of aquatic insects in several ways. Dragonfly populations are relatively small compared to other aquatic inver­ tebrates, and emergence of adult dragonflies is usually over a period of weeks or even a month or more (Corbet 1962). Adults are long-lived (up to 10 weeks; Corbet 1980) and reproductively active for a large portion of that time, and flight occurs primarily during dayli ht hours. Consequently, mor­ tality from collisions with vehicles may affe populations more se­ verely than other aquatic insect groups. Couple the destruction and degradation of wetlands that is occurring in many regions, the effect of road mortality on small or disjunct populations of dragonflies could be substan­ tial. I investigated the extent, species composition and sex ratio of road-killed adult dragonflies at a freshwater coastal wetland along the northern shore of Lake Huron. A major two-lane highway borders the wetland, separating the upland forest from the wetland itself. Because dragonflies, like most aquatic insects, depend on terrestrial habitat during the adult part of their life cycle (Wallace and Anderson 1996), they must cross the highway repeatedly to ac­ cess the different parts of their required habitat. This situation is ideal for investigating road mortality in dragonflies and collecting information which will be important for guiding wetland management plans and directing fu­ ture research.

MATERIALS A.1\fD METHODS Study Area. Mackinac Bay is located between the towns of Hessel and Cedarville in the eastern Upper Peninsula of Michigan immediately adjacent to a two-laned, paved highway (M-134). Mackinac Bay contains an extensive coastal wetland typi cal of the region. The vegetation of the wet meadow por­ tion of the wetland was predominately tussocks of Carex stricta, Carex aquatilis and Calamagrostis canadensis interspersed with pockets of stand­ ing water and shrubs. The wet meadow then graded into homogenous stands (without tussocks) of Carex lasiocarpus. Scirpus acutus interspersed with patches of Typha angustifolia (D. M. Albert and T. M. Burton, unpublished data), Pontederia cordata, and floating vegetation (Nuphar and Poto­ nwgeton sp.) dominated the deeper-water portions of the Two small streams run through the marsh and empty into Mackinac Bay. The road (M-13'i) was a two-Ianed, asphalt highway leI to the northern coast of Lake Huron in the ea Michigan. It was adjacent to the coastal wetland at Mack Bay. The high­ way ran parallel to the junction of the wet meadow and the forested part of the wetland such that it was bordered on the north by cedar and tamarack forest and to the south by the wetland. Traffic volume during 1997 along this portion of M-134 was estimated at 5743 vehicles per day (Michigan Depart­ ment of Transportation 1998, unpublished data), Dragonfly Sampling. I collected adult dragonflies which had been killed along M-134 at Mackinac Bay by walking a marked, 500 m segment of the

Published by ValpoScholar, 1999 3 The Great Lakes Entomologist, Vol. 32, No. 1 [1999], Art. 9

66 THE GREAT LAKES ENTOMOLOGIST Vol. 32, No. 1 & 2

highway which directly bordered the wetland. During each I walked up one side of the road and down the other. I collected dragonfly corpses off the highway, off the gravel verge, and out of the bordering grass. I collected corpses on 26 different days between 27 June 1997 and 8 August 1997. Be­ cause I did not sample throughout the entire summer, early and late summer species were under-represented. All collecting was done between 1700 and 1900 EST. Dragonflies were identified to species level and sexed to the extent that the condition of the specimen allowed. Identifications were made using keys (Walker 1953, Needham and Westfall 1955, Walker 1958, Walker and Corbet 1975) or by comparison with reference specimens. Statistical Analysis. I calculated daily mortality rate as the mean num­ ber of corpses collected per sampling day per kilometer for each species. The section of the highway used for collecting was 500 m; thus, daily mortality was equal to twice the number of corpses collected on a particular day. Be­ cause dragonfly densities are not constant, but rather peak towards the mid­ dle of the adult flight period, I also calculated "peak daily mortality rate." This value was the mean of the three sampling days with the highest num­ ber of casualties and estimated the maximum rate of mortality at each species' peak density. It is very possible that my sampling protocol produced overestimates of mortality rates because I did not always sample on consecutive days. Munguira and Thomas (1992) found that butterfly corpses remained on the road verge for over two weeks, and for some sampling events in my study, more than 24 hours had elapsed since my previous sampling event. Thus, corpse counts from those days may have represented the cumulative mortal­ ity over several days since the previous sampling event which would result in overestimates of actual daily mortality. On two occasions, I left 10 marked specimens of Aeshna canadensis Walker and Dorocordulia libera (Selys) on the highway verge, and, on both occasions, none of the 10 specimens were found 24 hours later. Also, estimates of mortality from days when> 24 hours had elapsed since the previous collection were not significantly greater than estimates from sampling events when only 24 hours had elapsed. This evi­ dence suggests that longevity of corpses did not cause overestimates of daily mortality rates in this study. It is also possible, however, that I have underestimated mortality for two reasons. First, the dry, sunny conditions along the highway verge resulted in corpses which dried in a matter of hours. These dried corpses were very light, and many were likely blown off the verge by coastal vv'inds which were com­ mon (S. Riffell, personal observation). These corpses would not have been col­ lected during the sampling events. Second, although Munguira and Thomas (1992) indicated that very few, if any, butterfly corpses were carried off in ra­ diators or grills, my observations at Mackinac Bay suggest that at least some dragonfly corpses were removed from the study area in this manner. I calculated the sex ratio of casualties as the percent of the corpses that were male. Corpses too damaged to confidently determine sex were not in­ cluded in this calculation. To test if sex ratios among casualties were differ­ ent from 1.00 (i.e. 50% male, 50% female), I used a logarithm-based G-statis­ tic (Zar 1984). G-statistics were not calculated if <10 specimens of a particular species or taxonomic group were collected.

RESULTS I collected a total of 1140 dragonfly corpses along the highway bordering the Mackinac Bay wetland. Corpses were from seven different families and

https://scholar.valpo.edu/tgle/vol32/iss1/9 4 ==~------~------~-- Riffell: Road Mortality of Dragonflies (Odonata) in a Great Lakes Coastal

1999 THE GREAT LAKES ENTOMOLOGIST 67

1.0

0.8 1 I ~ 0.6

0.2

0.0 Gomphus Macromia Somatochlora Leucorrhinia Aeshna Dorocordu/ia Epitheca Libellula

Figure 1. Sex ratios (% male) of road casualties at Mackinac Bay for selected genera ofdragonflies. Error bars represent ± 1 SE (according to Zar 1984).

25 different species. Just over two-thirds of the corpses were comprised of just four species: Libellula quadrimaculata Linn., Aeshna canadenesis, Ep­ itheca spinigera Selys, and Dorocordulia libera. Daily Mortality Rates. Mean daily mortality of all Odonata was 87.69 dragonfly casualties per kilometer. The highest number of casualties on a single day was 256 corpseslkm; the lowest was 10 corpses/km (Appendix A). For individual species, Libellula quadrimaculata had the highest mean daily mortality rate (24.00 kmiday [Table 1, Appendix AD. For most species and higher taxonomic groupings, mortality rates were highly variable. Sex Ratio of Casualties. OveralL 73% of the road casualties were male (Appendix B), but the sex ratio varied among genera (Figure 1). Mortality was significantly male-skewed for Aeshna (82% male; P < 0.001), Epitheca (80%; P < 0.001), Leucorrhinia (85%; P < 0.001), and LibeZlula (72%; P < 0.001) while mortality of Somatochlora (30%; P < 0.050) was significantly fe­ male-skewed. Mortality of Macromia (59%); Dorocordulia (54%) and Gom­ phus (37%) were not significantly sex-skewed (Figure 1; Appendix B). At the species level, 20 of the 25 species collected had male-skewed mortality, and this was statistically significant for six species: Aeshna canadensis, Anax Ju­ nius (Drury), Epitheca spinigera, Leucorrhinia sp., Libellula Julia (Uhler), and L. quadrimaculata. Causalities were female-skewed in four species, but only one (Somatochlora walshi (Scudder» was significant (Appendix B).

DISCGSSION Daily Mortality Rates. Daily mortality rates of dragonflies at Mackinac Bay were larger than published rates for other invertebrate taxa (Table I), and agree with the high mortality rate of Anax Junius (Drury) reported by Beckemeyer (1996 and Table 1). Such high rates of mortality were not sur­

Published by ValpoScholar, 1999 5 The Great Lakes Entomologist, Vol. 32, No. 1 [1999], Art. 9

68 THE GREAT LAKES ENTOMOLOGIST Vol. 32, No. 1 & 2

prising because of the different habitats that dragonflies require during dif­ ferent stages of the adult lifespan. Immediately after emergence, adults of most species of dragonflies leave the wetland and go to nearby terrestrial habitats (forests, meadows), where they shelter and feed (Corbet 1962). When they reach reproductive maturity (typically a few weeks; Corbet 1980), they return to the wetland to defend territories, copulate, and oviposit. At Mackinac Bay, this highway separates the terrestrial forest almost com­ pletely from the wetland, and nearly all movement between the two habitats requires that dragonflies cross the highway. Such positioning of the highway maximizes the potential for dragonfly crossings and, hence, fatal collisions. Others have described a similar phenomenon during amphibian breeding migrations (van Gelder 1973, Farhig et a1. 1995), where an alarming percent­ age of a population may be killed when the amphibians must cross roads to reach their breeding habitats (van Gelder 1973). Dragonflies differ from am­ phibians, however, in that return trips for breeding are undertaken daily, or more often by some individuals. For dragonflies, the potential for road mor­ tality exists throughout the reproductive period of the organism, not just the initial migration period. Biologists and planners alike should consider the ef­ fects of road mortality on dragonfly populations and the effects of road posi­ tion on mortality rates. For small or isolated populations, such high rates of mortality could potentially threaten population persistence, and further re­ search about the effects of road mortality on population dynamics is sorely needed to accurately gauge this threat. My estimates of daily mortality were high, but extremely variable from day to day (see Appendix A), which reflects the wide array of factors which can impact the actual number of collisions that occur. In theory, the number of insect-car collisions could be influenced by (1) dragonfly density and be­ havior and (2) traffic density and speed (van Gelder 1973, Fahrig et a1. 1995). Both dragonfly and traffic characteristics can be influenced by a myriad of secondary factors. For instance, dragonfly density can be influenced by the amount and'quality of the wetland habitat (e.g. presence of fish decrease Odonate abundance; Batzer and Wissinger 1996), the species composition of dragonfly community (i.e. some species may be more densely populated), and dragonfly activity level which is in turn a function of weather, temperature, and behavior. Traffic density and speed could be influenced by time of day (e.g. work traffic), time of week (e.g. recreational weekend traffic), visibility, characteristics of the road (e.g. paved vs. nonpaved) and the size and charac­ teristics of the communities connected by the roadway in question. Unfortu­ nately, little is known about many ofthese factors, so predicting even relative mortality rates in other locations is not possible. Research about how these factors interact to influence road mortality rates in a variety of habitats and settings is necessary to accurately understand the effects of road mortality on dragonfly populations. Although the most abundant species generally had the highest mortality rates (S. Riffell, personal observation), quantitative data about the back­ ground densities of adult dragonflies at Mackinac Bay was not collected. Thus, it is not possible to determine the percent of total mortality of adult dragonflies that was caused by vehicular collisions, or to make any definitive conclusions about what effects road mortality could have on population dy­ namics. Also, one cannot make inferences about the relative susceptibilities of particular species or taxonomic groups. Some species, such as Libellula julia and Gomphus sp., were often seen basking on the road itself, and it would seem that they would suffer greater per capita mortality than other species which did not exhibit such behavior, but such data are lacking. Fu­ ture studies could use visual estimates of background population size (e.g.

https://scholar.valpo.edu/tgle/vol32/iss1/9 6 Riffell: Road Mortality of Dragonflies (Odonata) in a Great Lakes Coastal

1999 THE GREAT LAKES ENTOMOlOGIST 69

transect counts), make detailed observations of flight behavior andJor road crossing success, or mark individuals and estimate mortality as percentage ofmarked individuals among the casualties. Sex ratios of Casualties. Although males usually comprise only 40 - 50% of the population at emergence (Corbet 1980), over 70% of all dragonfly corpses were males (Appendix B). This is strong evidence that road mortality at Mackinac Bay was truly male-specific. However, it would be premature to conclude that males are actually more susceptible while crossing roads than are females. In most Odonata, males spend most or all of their reproductive lives at the wetland breeding sites defending territories, whereas females spend far less time at breeding sites (Corbet 1962; Hamilton and Mont­ gomerie 1989), often being absent for several days after successful oviposi­ tion (Michiels and Dhondt 1991). At the Mackinac Bay wetland, males would· be required to make at least two highway crossings per day, and females far fewer. Male-skewed mortality at Mackinac is at least partly due to the greater number of males crossing the highway and the hi er crossing fre­ quency per male. However, whether or not males have a ent probability of survival per crossing than females is not known. Estimating sex differ­ ences in crossing survival would require detailed observations of crossing events or recovery ofmarked individuals. Conservationists might conclude that road mortality would have little ef­ fect on the reproductive output of a population because most of the casualties were males. However, this generalization would be unwise because the sex ratios of casualties were even or actually female skewed for some species (e.g. Somatochlora sp., Figure 1 and Appendix B). Qualitatively, mortality of the most abundant species at the wetland (e.g. A. canadensis, L. quadrimac­ ulata, Leucorrhinia sp.; based on larval abundances; T. Burton, unpublished data) was the most heavily male-skewed, while mortality of less abundant species (e.g. Gomphus spicatus Hagen, Macromia illinioensis Walsh, Soma­ tochlora sp.) was less male-skewed or even female-skewed. Two mechanisms may account for this pattern. First, increased abundance of males may increase the number of territorial disputes among males, thus increasing their road crossing frequency (S. Riffell, personal observation). This would increase male mortality at high densities. Second, lower densities of males decreases the probability that a female can breed successfully (Michiels and Dhondt 1991), and females would have to make additional road-crossings until they are successful. This would increase female mortal­ ity at low densities. Although these ideas are speculative and based on quali­ tative observations, I have presented them for the sake of fostering further discussion and research. Sex differences in road mortality among species could also be a result of differences in the sex ratio of individuals at the wet­ land, differential flight abilities between sexes in some species but not oth­ ers, or other factors. Identifying causes of sex-skewed mortality is important because female-specific mortality could have a severe impact on reproductive output, and thus population persistence, for species with small andJor iso­ lated populations. Implications for Conservation. It is clear that large numbers of a wide array of dragonfly species can potentially become casualties of vehicular col­ lisions. For small or isolated populations such as the rare Hine's Emerald (Somatochlora hineana Williamson), losses from road mortality take on greater importance because small populations are particularly susceptible to extinction. Additionally, road mortality of congeneric Somatochlora species was female-skewed at Mackinac Bay, and the impact of female-skewed mor­ tality on populations of Hine's Emerald has not be investigated. Predicting the effects of road mortality on dragonfly populations and conserving road­

Published by ValpoScholar, 1999 7 The Great Lakes Entomologist, Vol. 32, No. 1 [1999], Art. 9

70 THE GREAT LAKES ENTOMOLOGIST Vol. 32, No.1 & 2

impacted populations will be difficult without further understanding of the causes and effects of road mortality in dragonflies. Priorities for research should focus on: (1) determining the percent of total adult mortality that is caused by vehicular collisions; (2) identifying the factors which affect cross­ ing survivorship and mortality rates; and (3) determining the effects of sex­ skewed mortality on the sex composition and reproductive potential of drag­ onfly populations. Dragonfly road mortality would be minimized if no additional roads were constructed in or near dragonfly habitats. When roads must be constructed, careful consideration of road placement could ameliorate potential impacts on dragonfly populations. Roads which are placed in critical breeding habitat (i.e. actual wetland) or which separate different habitats used by dragonflies (i.e. splitting forest and wetland) should not be constructed if reasonable al­ ternatives exist. Relative to other flying insects, dragonflies are extremely quick and agile flyers (Brodsky 1994). It is very possible that merely reduc­ ing traffic speeds in critical areas may significantly reduce insect-vehicle col­ lisions, but this idea has not been tested. Other options include constructing temporary tunnels over sections of roads with high mortality to prevent dragonflies from crossing at car level or resting on the road surface. More re­ search should be conducted, however, to determine the true impact of road mortality on dragonfly populations and to assess the effectiveness of pro­ posed management options before time- and resource-consuming conserva­ tion efforts are implemented.

ACKNOWLEDGMENTS This research was conducted while the author was supported by a grant from the Michigan Chapter of the Nature Conservancy to T. Burton. T. Cashatt, M. O'Brien, W. Steffens, and K. Tennessen helped by verifying spec­ imens. This paper benefited from reviews by T. Burton, M. Kielb, R. Merritt, A. Riffell, and an anonymous reviewer. Reference specimens were deposited in the University ofMichigan Museum of Zoology, Insect Division.

LITERATURE CITED Ashley, E. P. and J. T. Robinson. 1996. Road mortality of amphibians, reptiles, and other wildlife on the Long Point Causeway, Lake Erie, Ontario. Canadian Field-Nat­ uralist 110:403-412. Batzer, D. P. and S. A Wissinger. 1996. Ecology of insect communities in nontidal wet­ lands. Annu. Rev. Entomol. 41:75-100. Beckemeyer, R. 1996. Dragonflies as roadkill. Argia 8:19-20. Brodsky, A K. 1994. The evolution of insect flight. Oxford Univ., New York. 229 pp. Corbet, P. S. 1962. A biology of dragonflies. H. F. and G. Witherby Ltd., London. 247 pp. Corbet, P. S. 1980. Biology of odonata. Annu. Rev. Entomol. 25:189-217. Doak, D. F. 1995. Source-sink models and the problem of habitat degradation: general models and application to the Yellowstone Grizzly. Conservation BioL 9:1370-1379. Edmunds, Jr., G. F. and R. D. Waltz. 1996. Ephemeroptera, pp. 126-163. In: R.W. Mer­ ritt and K. W. Cummins (eds.), Aquatic insects of North America. KendalllHunt Pub­ lishing, Dubuque, Iowa. 862 pp. L., J. H. Pedlar, S. E. Pope, P. D. Taylor and J. F. Wegner. 1995. Effect of road on amphibian density. BioI. Conservation 73:177-182. Forman, R. T. T. 1995. Land mosaics: the ecology of landscapes and regions. Cam­ bridge Univ., New York.

https://scholar.valpo.edu/tgle/vol32/iss1/9 8 Riffell: Road Mortality of Dragonflies (Odonata) in a Great Lakes Coastal

1999 THE GREAT LAKES ENTOMOLOGIST 71

Hamilton, L. D., and R. D. Montgomerie. 1989. Population demography and sex ratio in a Neotropical damselfly (Odonata: Coenagrionidae) in Costa Rica. Jour. Trop. Ecol. 5:159-171. Harris, L. D. and P. B. Gallagher. 1989. New initiatives for wildlife conservation: the need for movement corridors, pp.11-34. In: G. Mackintosh (ed.), Preserving commu­ nities and corridors. Defenders of Wildlife, Washington, D. C. 96 pp. Lalo, J. 1987. The problem of road-kill. Am. Forests (Sept-Oct.): 50-52,72. Merriam, G., M. Kozakiewicz, E. Tsuchiya and K. Hawley. 1989. Barriers as bound­ aries for metapopulations and demes of Peromyscus leucopus. Landscape Eco!. 2:227-235. Michiels, N. K. and A. A. Dhondt. 1991. Sources of vmiation in male mating success and female oviposition rate in a nonterritorial dragonfly. Behavioral EcoL and Socio­ bioI. 29:17-25. Munguira, M. L. and J. A. Thomas. 1992. Use of road verges by butterfly and burnet populations, and the effect of roads on adult dispersal and mortality. Jour. Appl. Ecol. 29:316-329. Needham, ,1. G. and M. J. Westfall. 1955. A manual of the dragonflies of North America (Anisoptera). Univ. Calif., Berkeley. 615 pp. Rosen, P. C. and C. H. Lowe. 1994. Highway mortality of snakes in the Sonoran Desert of southern Arizona. BioI. Conservation 68:143-148. Seibert, H. C. and J. H. Conover. 1991. Mortality of vertebrates and invertebrates on an Athens County, Ohio, highway. Ohio Jour. Sci. 91:163-166. van Gelder, J. J. 1973. A quantitative approach to the mortality resulting from traffic in a population of Bufo bufo. Oecologia 13:93-95 . .Walker,E. M. 1953. The Odonata of Canada and Alaska, Vol.l. Univ. Toronto, Toronto. 292 pp. Walker, E. M. 1958. The Odonata of Canada and Alaska, Vol. 2. Univ. Toronto, '!bronto. 318 Walker, M. and P. S. Corbet 1975. The Odonata of Canada and Alaska, Vol. 3. Univ. Toronto, Toronto. 307 pp. Wallace, J. B. and N. H. Anderson. 1996. Habitat, life history, and behavioral adapta­ tions of aquatic insects, pp. 41-73. In: R. W. Merritt and K. W. Cummins (eds.), Aquatic insects of North America. KendalllHunt Publishing, Dubuque, Iowa. 862 pp. Wiggins, G. B. 1996. Trichoptera families, pp. 309-349. In: R. W. Merritt and K. W. Cummins (eds.), Aquatic insects of North America. KendalllHunt Publishing, Dubuque, Iowa. 862 pp. Zar, J. H. 1984. Biostatistical analysis. Prentice-Hall, Englewood Cliffs, New Jersey. 718 pp.

Appendix A. Mean daily road-mortality ratos (# of corpses) and peak mortality rates (per kilometer per day) of adult dragonflies at Mackinac Bay, Michigan dur­ ing summer 1997. Peak Mortality Daily Mortality Rates Rates* Species Mean ± 1 SD Range Mean 1 SD CALOPTERYGIDAE Calopteryx aequabilis Say 0.23 0.65 2.0·-0.0 2.00 ± 0.00 Calopteryx rnaculata (Beauvois) 0.46 ± 1.03 4.0-0.0 2.67 ± 1.15

CORDULEGASTRIDiill Cordulegaster maculatus Selys 0.23 ± 0.65 2.0-0.0 2.00 ± 0.00 Continued

Published by ValpoScholar, 1999 9 ------The Great Lakes Entomologist, Vol. 32, No. 1 [1999], Art. 9

72 THE GREAT LAKES ENTOMOLOGIST Vol. 32, No. 1 &2

AppendixA. Continued. Peak Mortality Daily Mortality Rates Rates* Mean ± 1 SD Range Mean 1 SD

GOMPHIDAE Arigomphus cornutus Tough 0.54 ± 1.33 4.0-0.0 4.00 0.00 Gomphus spicatus Hagen 1.46 ± 2.56 8.0-0.0 6.67 ± 1.15 Unk. Gomphidae 1.85 ± 4.19 18.0-0.0 12.00 ± 5.29 Total Gomphidae 3.85 ± 5.42 18.0-0.0 14.67 3.06

AESHNIDAE Aeshna canadensis Walker 12.15 ± 16.72 60.0-0.0 48.00 ± 12.00 Aeshna interrupta Walker 0.38 ± 1.13 4.0-0.0 3.33 1.15 Aeshna eremita Scudder 0.15 ± 0.54 2.0-0.0 2.00 0.00 Aeshna umbrosa Walker 0.15 ± 0.78 4.0-0.0 4.00 ** Unk. Aeshna sp. 1.77 ± 3.11 14.0-0.0 8.67 5.03 ThtalAeshna sp. 14.62 ± 19.48 70.0-0.0 58.67 ± 10.26 Anaxjunius (Drury) 1.62 ± 1.79 6.0-0.0 5.33 ± 1.15 Basiaeschnajanata (Say) 0.31 ± 0.74 2.0-0.0 2.00± 0.00 Gomphaeschna furcillata (Hagen) 0.46 ± 1.17 4.0-0.0 3.33 ± 1.15 Total Aeshnidae 17.00 ± 18.85 72.0-0.0 60.00 ± 11.14

MACROMIIDAE Macromia illinoiensis Walsh 1.54 ± 2.85 12.0-0.0 8.00 ± 3.46

CORDULIIDAE Cordulia shurtleffi Scudder 0.46 ± 1.03 4.0-0.0 2.67 ± 1.15 Dorocordulia Libera (Selys) 6.77 7.65 24.0-0.0 22.67 ± 1.15 Somatochlora franklini (Selys) 0.08 ± 0.39 2.0-0.0 2.00 ± ** Somatochlora walshi (Scudder) 1.31 ± 1.95 6.0-0.0 5.33 ± L15 Somatochlora williamsoni Walker 0.38 ± 0.98 4.0-0.0 2.67 ± 1.15 Unk. Somatochlora sp. 0.38 ± 0.98 4.0-0.0 2.67 ± 1.15 Total Somatochlora sp. 2.15 ± 3.15 12.0-0.0 9.33 ± 3.06 Epitheca spinigera Selys 15.31 ± 17.44 70.0-0.0 50.67 ± 16.77 Unk. Epitheca sp. 1.62 ± 3.87 14.0-0.0 11.33 ± 3.06 Total Corduliidae 27.85 ± 28.05 102.0-0.0 87.33 ± 18.90

LIBELLULIDAE Leucorrhinia sp. 4.15 ± 6.14 18.0-0.0 16.67 ± 1.15 Libellula julia (Uhler) 8.15 ± 13.05 58.0-0.0 36.00 ± 19.08 LibelluLa pulchella Drury 1.54 ± 2.14 8.0-0.0 5.33 ± 2.31 Libellula quadrimaculata Linn. 24.00 ± 24.57 92.0-0.0 69.33 ± 19.63 Total Libellula sp. 33.69 ± 37.65 154.0-0.0 107.33 ± 40.46 Sympetrum corruptum (Hagen) 0.08 ± 0.39 2.0-0.0 2.00 ± ** Sympetrum obtrusum (Hagen) 0.15 0.54 2.0-0.0 2.00 ± 0.00 Total Libel1ulidae 38.08 ± 41.37 164.0-0.0 121.33 ± 37.00

TOTAL ODONATA 87.69 68.41 256.0-10.0 226.00 ± 26.15 * Peak rates are the mean of the three dates with the highest mortality for that taxo­ nomic grouping. ** Standard deviations are not provided for species that were only collected on a single date.

https://scholar.valpo.edu/tgle/vol32/iss1/9 10 Riffell: Road Mortality of Dragonflies (Odonata) in a Great Lakes Coastal

1999 THE GREAT LAKES ENTOMOLOGIST 73

Appendix B. Sex ratios (% male) of dragonfly casualties at the Mackinac Bay wetland during summer 1997. % G- P- Species Males Females Unknown Total Male statistic* value CALOPTERYGIDAE Calopteryx aequabilis 3 0 0 3 100 Calopteryx maculata 1 4 1 6 17 CORDULEGASTRIDAE Cordulegaster maculatus 3 0 0 3 100 GOMPHIDAE Arigomphus cornutus 6 1 0 7 86 Gomphus spicatus 7 12 0 19 37 1.331 NS Unk. Gomphidae 12 7 5 24 63 1.331 NS Thtal Gomphidae 25 20 5 50 56 0.557 NS AESHNIDAE Aeshna canadensis 129 25 4 158 84 76.883 <0.001 Aeshna interrupta 4 1 0 5 80 Aeshna eremita 2 0 0 2 100 Aeshna umbrosa 2 0 0 2 100 Unk. Aeshna sp. 12 7 4 23 63 1.331 NS Total Aeshna sp. 149 33 8 190 82 79.993 <0.001 18 Anaxjunius 18 3 0 21 86 11.887 <0.001 Basiaeschna janata 3 1 0 4 75 Gomphaeschna furcillata 4 2 0 6 67 Total Aeshnidae 174 39 8 216 82 92.479 <0.001 MACROMIIDAE Macromia illinoiensis 10 7 3 20 59 0.532 NS CORDULlIDAE Cordulia shurtleffi 3 2 1 6 60 Dorocordulia Libera 44 37 7 88 54 0.606 NS Somatochlora franklini 1 0 0 1 100 Somatochlora walshi 5 12 0 17 29 2.970 <0.100 Somatochlora williamsoni 1 4 0 5 20 Unk. Somatochlora sp. 1 3 1 5 25 Total Somatochlora sp. 8 19 1 28 30 4.615 <0.050 Epitheca spinigera 152 38 9 199 80 73.243 <0.001 Unk. Epitheca sp. 12 3 6 21 80 5.782 <0.025 Thtal Corduliidae 229 106 27 362 68 46.235 <0.001 LIBELLULIDAE Leucorrhinia sp. 44 8 2 54 85 27.438 <0.001 Libellula julia 71 22 13 106 76 27.168 <0.001 Libellula pulchella 10 7 3 20 59 0.532 NS Libellula quadrimaculata 191 79 42 312 71 47.893 <0.001 Total Libellula sp. 272 108 58 438 72 73.158 <0.001 Sympetrum corruptum 1 0 0 1 100 Sympetrum obtrusum 1 1 0 2 50 Thtal Libellulidae 318 117 60 495 73 96.500 <0.001

TOTAL ODONATA 753 286 101 1140 73 217.612 <0.001 '" Zar (1984). Statistical hypotheses were not tested if < 10 individuals were collected.

Published by ValpoScholar, 1999 11